Late Cretaceous to Early Paleocene Climate and Sea-Level £Uctuations: the Tunisian Record

Home , Danian, Late Cretaceous, Paleocene

Published in Palaeogeography, Palaeoclimatology, Palaeoecology 2754 : 1-32, 2002, 1 which should be used for any reference to this work

Late Cretaceous to early Paleocene climate and sea-level £uctuations: the Tunisian record

Thierry Adatte a;*, Gerta Keller b, Wolfgang Stinnesbeck c

a Institut de Ge¨ologie, 11 Rue Emile Argand, 2007 Neuchatel, Switzerland b Department of Geosciences, Princeton University, Princeton, NJ 08544, USA c Geologisches Institut, Universitat Karlsruhe, 76128 Karlsruhe, Germany

Abstract

Climate and sea-level fluctuations across the Cretaceous^Tertiary (K^T) transition in Tunisia were examined based on bulk rock and clay mineralogies, biostratigraphy and lithology in five sections (El Melah, El Kef, Elles, Ain Settara and Seldja) spanning from open marine to shallow inner neritic environments. Late Campanian to early Danian trends examined at El Kef and Elles indicate an increasingly more humid climate associated with sea-level fluctuations and increased detrital influx that culminates at the K^T transition. This long-term trend in increasing humidity and runoff in the Tethys region is associated with middle and high latitude cooling. Results of short-term changes across the K^T transition indicate a sea-level lowstand in the latest Maastrichtian about 25^100 ka below the K^T boundary with the regression marked by increased detrital influx at El Kef and Elles and a short hiatus at Ain Settara. A rising sea-level at the end of the Maastrichtian is expressed at Elles and El Kef by deposition of a foraminiferal packstone. A flooding surface and condensed sedimentation mark the K^T boundary clay which is rich in terrestrial organic matter. The P0^ P1a transition is marked by a sea-level lowstand corresponding to a short hiatus at Ain Settara where most of P0 is missing and a period of non-deposition and erosion in the lower part of P1a (64.95 Ma). At Seldja, P0 and possibly the topmost part of CF1 are missing. These sea-level fluctuations are associated with maximum humidity. These data suggest that in Tunisia, long-term environmental stresses during the last 500 ka before the K^T boundary and continuing into the early Danian are primarily related to climate and sea-level fluctuations. Within this long-term climatic trend the pronounced warm and humid event within the latest Maastrichtian Zone CF1 may be linked to greenhouse conditions induced by Deccan volcanism. The absence of any significant clay mineral variations at or near the K^T boundary and Ir anomaly suggests that the bolide impact had a relatively incidental short-term effect on climate in the Tethys region.

Keywords: sea-level; climate £uctuations; K^T boundary; Upper Campanian; Maastrichtian; Tunisia; bulk; clay minerals; organic matter; geochemistry

1. Introduction

There are relatively few studies that detail long- * Corresponding author. Tel.: +41-32-718-26-18; Fax: +41- 32-718-26-01. term climate and sea-level changes during the late E-mail address: [email protected] (T. Adatte). Cretaceous which was generally assumed to have 2

been equally warm. Recent stable isotope studies, addition, climatic changes inferred from clay min- however, have revealed that the Maastrichtian eral contents correlate with sea-level changes. global climate was signi¢cantly cooler than during Warm or humid climates accompany high sea-lev- the earlier Cretaceous. Strong climate and temper- els and cooler or arid climates generally accom- ature £uctuations mark the late Campanian and pany low sea-level (Li et al., 2000). Maastrichtian, as indicated from stable isotope The global sea-level £uctuations are linked to records from the equatorial Paci¢c (Site 463, Kel- climatic changes (Li et al., 1999) and inversely ler and Li, in press) and middle and high latitude correlate with species diversity in planktic fora- South Atlantic (Sites 525, 689 and 690, Barrera minifera (e.g. diversity maximum follows maxi- and Huber, 1990; Barrera, 1994; Barrera et al., mum cooling at 70.7^70.3 Ma; diversity decline 1997; Li and Keller, 1998a,b). The ¢rst major follows warming at 65.4^65.2 Ma, Li and Keller, global cooling occurred between 71 and 73 Ma 1998a,c). But precisely how climate changed and decreased intermediate water temperatures across the K^T boundary mass extinction and by 5^6³C and surface temperatures by 4^5³C in during the subsequent evolution of early Tertiary middle and high latitudes. Between 68.5 and 70 faunas is still an enigma largely because diagenetic Ma, intermediate waters warmed by 2³C. Global alteration of carbonates obscures the oxygen iso- cooling resumed between 68.5 and 65.5 Ma when tope records (see Stueben et al., this volume). intermediate water temperatures decreased by 3^ A number of studies have attempted to recon- 4³C and sea surface temperatures decreased by struct the sea-level history across the K^T transi- 5³C in middle latitudes. About 450^200 ka before tion in Tunisia based on benthic and planktic for- the Cretaceous^Tertiary (K^T) boundary rapid aminifera, dino£agellates or palyno£oras (e.g. global warming increased intermediate and sea Brinkhuis and Zachariasse, 1988; Keller, 1988b, surface temperatures by 3^4³C, though sea sur- 1992; MacLeod and Keller, 199la,b; Schmitz et face temperatures changed little in low latitudes al., 1992; Speijer, 1994; Keller and Stinnesbeck, (Li and Keller, 1998b). Beginning about 200 ka 1996; Brinkhuis and Visscher, 1994; Brinkhuis et before the end of the Maastrichtian, climate al., 1998; Galeotti and Coccioni, 2002). The over- cooled rapidly by 2^3³C in both surface and in- all sea-level trends in these studies are in general termediate waters and warmed again during the agreement, though may vary in the details and the last 50^100 ka of the Maastrichtian (Li and Kel- timing of sea-level lowstands. A major problem ler, 1998b; Stueben et al., this volume). has been the absence of oxygen isotope-inferred These global climatic changes were associated climate data to support sea-level £uctuations in- with major sea-level £uctuations as expressed in ferred from faunal and £oral proxies. Oxygen iso- a variety of sea-level proxies, including bulk rock tope data are frequently not reliable temperature and clay mineral compositions, stable isotopes, indicators across the K^T transition because of total organic carbon (TOC), Sr/Ca ratios and diagenetic alteration of carbonate and recrystalli- macro- and microfaunal associations. Based on zation of foraminiferal tests (Oberhaensli et al., these sea-level and climate proxies, Li et al. 1998; Stueben et al., this volume). Alternate tem- (1999, 2000) have identi¢ed seven major sea-level perature proxies based on the coiling direction of regressions during the last 10 myr of the Creta- the benthic foraminifera Cibicidoides pseudoacutus ceous at El Kef and Elles (Tunisia): late Campa- (Galeotti and Coccioni, 2002) and the ratio of nian (V74.2 Ma, 73.4^72.5 Ma and 72.2^71.7 warm/cool dinocysts (Brinkhuis et al., 1998) Ma), early Maastrichtian (70.7^70.3 Ma, 69.6^ have been proposed, but these have yet to be in- 69.3 Ma and 68.9^68.3 Ma), and late Maastrich- dependently con¢rmed. tian (65.45^65.50 Ma). Low sea-levels are gener- This study evaluates climate and sea-level ally associated with increased terrigenous in£ux, changes across the K^T transition based on litho- low kaolinite/chlorite+illite ratios, high TOC and logical characteristics, bulk rock and clay mineral high Sr/Ca ratios, whereas high sea-levels are gen- data from several Tunisian sections that span erally associated with the reverse conditions. In from upper slope (El Melah) to outer neritic (El 3

Fig. 1. Paleoenvironmental settings of ¢ve Tunisian K^T sections spanning from the restricted shallow Gafsa Basin (Seldja section) at the edge of the Sahara to the middle and outer shelf depths of the El Kef, Elles and Ain Settara sections to the north of the Kasserine Island, and to the upper bathyal El Melah section to the north (modi¢ed after Burollet, 1956; Burollet and Oudin, 1980). Isopach lines are given in meter for the Maastrichtian interval.

Kef, Elles), middle neritic (Ain Settara) and inner trichtian of Elles and El Kef sections. Biostrati- neritic (Seldja) environments (Fig. 1). The time graphic data for each section are based on pub- interval analyzed spans from the uppermost lished studies: Elles I and El Melah from Karoui- Maastrichtian Zone CF1 to the early Danian Yakoub et al. (2002), Elles II K^T transition from Zone Plb or Plc. In addition, we examine long- Keller et al. (2002) and upper Maastrichtian from term trends based on bulk rock and clay mineral Abramovich and Keller (2002), Ain Settara from data from the late Campanian through Maas- Luciana (2002) and Seldja from Keller et al. trichtian of the Elles and El Kef sections. (1998). Sediment accumulation rates were calculated based on the time scale of Cande and Kent (1995). 2. Methods Whole rock and clay mineral analyses were conducted at the Geological Institute of the Uni- In the ¢eld, sections were cleaned from surface versity of Neuchatel, Switzerland, based on XRD contamination by digging a trench to fresh bed- analyses (SCINTAG XRD 2000 Di¡ractometer). rock. Samples were then collected at 5^10 cm in- Sample processing followed the procedure out- tervals and at closer 1^2 cm intervals across the lined by Ku«bler (1987) and Adatte et al. (1996). K^T boundary clay layer. For each section, the XRD analyses of the whole rock were carried out same sample set was used for faunal, geochemical for all the samples at the Geological Institute of and mineralogical studies to insure direct compar- the University of Neuchatel. The samples were ison of results (see Keller et al., 2002; Stueben et prepared following the procedure of Ku«bler al., this volume). Biostratigraphy is based on (1987). Random powder of the bulk samples is planktic foraminifera and the zonation of Keller used for characterization of the whole rock min- et al. (1995) for the K^T transition and of Li and eralogy. Nearly 20 g of each rock sample was Keller (1998c) for the late Campanian and Maas- ground with a `jaw' crusher to obtain small rock 4

chips (1^5 mm). Approximately 5 g was dried at a Determination of chlorite and kaolinite is ob- temperature of 60³C and then ground again to a tained by deconvolution of their 002 and 004 homogeneous powder with particle sizes 6 40 peaks respectively at 24.9³ (kaolinite) and 25.2³ Wm. About 800 mg of this powder was pressed (chlorite). (20 bar) in a powder holder covered with a blot- Organic carbon analysis was conducted using a ting paper and analyzed by XRD. Whole rock CHN Carlo-Erba Elemental Analyzer NA 1108. composition is based on methods described by Total carbon was ¢rst measured on bulk samples Ferrero (1965, 1966), Klug and Alexander (1974) (0.01^0.02 g). Total organic carbon (TOC) was and Ku«bler (1983). This method for semi-quanti- determined after removing carbonate by acidi¢ca- tative analysis of the bulk rock mineralogy (ob- tion with hydrochloric acid (10%), assuming that tained by XRD patterns of random powder sam- dissolved organic matter in ancient sediments is ples) used external standards with an error nearly absent. The obtained values were com- varying between 5 and 10% for the phyllosilicates pared with a standard reference sample. Analyti- and 5% for grain minerals. cal precision for a standard is þ 0.003% and re- Clay mineral analyses were based on methods producibility for the Maastrichtian samples is by Ku«bler (1987). Ground chips were mixed with 0.01% for bulk rocks (total carbon) and 0.02% de-ionized water (pH 7^8) and agitated. The car- for insoluble residues. bonate fraction was removed with the addition of HCl 10% (1.25 N) at room temperature for 20 min, or more until all the carbonate was dis- 3. Lithology solved. Ultrasonic disaggregation was accom- plished during 3 min intervals. The insoluble res- 3.1. El Melah K^T idue was washed and centrifuged (5^6 times) until a neutral suspension was obtained (pH 7^8). Sep- The El Melah section is located near the village aration of di¡erent grain size fractions ( 6 2 Wm of El Aouana about 150 km northeast of El Kef and 2^16 Wm) was obtained by the timed settling and 60 km northwest of Tunis (see Karoui-Ya- method based on Stokes law. The selected frac- koub et al., 2002 for directions). The El Melah tion was then pipetted onto a glass plate and air- outcrop is 10 m thick and located on the right dried at room temperature. XRD analysis of ori- side of the Oued el Maleyh. The section spans 2 ented clay samples was made after air drying at m of the uppermost Maastrichtian Zone CF1 and room temperature and ethylene-glycol solvated 8 m of Danian sediments (Fig. 2). The uppermost conditions. The intensities of selected XRD peaks 5 cm of the Maastrichtian gray marls are highly characterizing each clay mineral present in the bioturbated. The K^T boundary is marked by a size fraction (e.g. chlorite, illite, kaolinite, smec- thin rusty red layer similar to Elles and El Kef. A tite) were measured for a semi-quantitative esti- 20 cm thick dark organic-rich clay of Danian age mate of the proportion of clay minerals present overlies the red layer and is strongly bioturbated in the size fractions 6 2 Wm and 2^16 Wm (error (Fig. 3). The earliest Danian Zone P0 spans the þ 5%). Therefore, clay minerals are given in rela- ¢rst 10 cm of this dark clay layer, followed by 1.4 tive percent abundance without correction factors. m of Zone Pla (Karoui-Yakoub et al., 2002). Content in swelling (% smectite) is estimated by Gray shales and marls alternate upsection and using the method of Moore and Reynolds (1989). marls are increasingly more dominant in Zones

Fig. 2. Biostratigraphic correlation of the K^T transition in the ¢ve studied sections which span a distance of more than 450 km from the north to the south of Tunisia. Note that Zone P0 is most expanded (50 and 60 cm) at the El Kef and Elles sections, condensed (10 cm) at El Melah, mostly missing at Ain Settara due to a hiatus (only 2 cm present), and absent at Seldja due to a hiatus. 5 6

P1a^P1b. Beginning at 7.40 m above the K^T the west of El Kef. Sediment accumulation rates boundary (P1d) marly limestones are interbedded based on the time scale of Cande and Kent (1995) with marls (Fig. 2). average 2 cm/ka for the latest Maastrichtian Zone Paleogeographically, El Melah was located in CF1 (Li and Keller, 1998a,b,c), about 1.1 cm/ka an outer neritic to upper bathyal environment for Zone P0, and 1.9 cm/ka for Zone Pla. (300^400 m deep in the Tunisian trough about 200 km north of the Kasserine Island (Fig. 1). 3.3. Elles I and Elles II K^T The El Melah area was thus less in£uenced by terrigenous in£ux than sections to the south (Bur- The Elles sections are located in the Karma ollet, 1956). Consequently, sediment accumulation valley about 75 km southeast of El Kef near the rates are lower compared to El Kef and Elles. At hamlet of Elles (see Karoui-Yakoub et al., 2002 El Melah, sediment accumulation rates average for directions). Campanian through Eocene sedi- 0.35 cm/ka for the P0^P1a interval based on the ments outcrop along the Karma valley which time scale of Cande and Kent (1995). The com- forks into two parts within late Maastrichtian paratively low sediment accumulation rate mainly sediments. The Elles I section was collected in re£ects sediment starvation due to its paleo-loca- the right valley fork which exposes about 7^8 m tion and distance from a terrigenous source. of Maastrichtian sediments and about 20 m of Danian sediments (Fig. 2) (see Karoui-Yakoub 3.2. El Kef K^T et al, 2002). Elles II was collected in the left valley fork which exposes about 50 m of Maastrichtian The El Kef K^T stratotype is located in north- sediments and continues for several hundred me- western Tunisia about 7 km west from the town ters across the K^T boundary and through the of El Kef and 75 km from Elles (Fig. 1; see Keller Paleocene. et al., 1995 for precise location). In recent years, At Elles I, the latest Maastrichtian is composed the section has been heavily sampled and agricul- of about 7^8 m of gray shales and marls with tural encroachment and grazing obscure the out- several resistant marly limestone layers between crop (Remane et al., 1999). The uppermost Maas- 4 and 7 m below the K^T boundary. The K^T trichtian consists of 4.5 m of gray marls (CF1) transition is well marked by a 0.5^1.0 cm clay followed by a 3 mm thick rusty red clay layer layer and a 3^4 mm thick rusty red layer em- (Figs. 2, 3). This red layer is between two thin bedded between two gypsum-jarosite layers of layers of secondary gypsum, goethite and jarosite, late diagenetic origin. The red clay layer is over- similar to Elles and El Melah sections. About 20 lain by a 60 cm thick dark clayey interval (Zone cm below the K^T boundary red layer is a 10 cm P0), followed by 5.5 m of gradually lighter shales thick layer which is signi¢cantly enriched in fora- and shaley marls. Upsection, these sediments are minifera and correlates with the foraminiferal interbedded with marly limestones (P1b). The packstone observed at Elles I and II outcrops stratigraphic interval considered in this study in- (Fig. 3) (Keller et al., 2002). Above the K^T red cludes the uppermost meter of the Maastrichtian clay layer is a 60 cm thick dark gray shale layer (CF1 Zone), the K^T boundary (P0) and the ¢rst (Zone P0) which grades upwards into marly 7 m of the Danian (P1a^P1b, Figs. 2, 3) biostra- shales (Fig. 2, P1a^P1b). tigraphy after Karoui-Yakoub et al., 2002). Compared with El Melah, the El Kef section Elles II di¡ers from Elles I primarily in the was located closer to the emerged Kasserine Is- more expanded K^T transition and the presence land and in a shallower outer neritic to upper of a 5^10 cm foraminiferal packstone with ripple bathyal environment (V200^300 m, Fig. 1). Sedi- marks below the K^T boundary clay and red ment deposition occurred in open marine condi- layer (Fig. 3). The uppermost Maastrichtian at tions, but with signi¢cant terrigenous in£ux pre- Elles II is characterized by a monotonous se- dominantly from the Kasserine Island and quence of dark gray siltstones, silty shales, marls intermittent in£ux from emerged land located to and sandy marls (CF1). A few Fe concretions 7

Fig. 3. Expanded stratigraphic interval and lithological characteristics across the K^T transition in the ¢ve studied outcrops. With the exception of Seldja, all sections contain the characteristic thin red layer rich in Ir and spinels which marks the K^T boundary mass extinction of tropical and subtropical planktic foraminifera. Dark organic-rich clay overlies the red layer in all sections and marks Zone P0. Gypsum of diagenetic origin is usually present above and below the red layer. At Elles II a second red layer is present. Hiatuses at or near the K^T boundary can be identi¢ed by the thin Zone P0 clay layer (Ain Settara) and the truncated burrows (El Melah, Elles II, Ain Settara (two intervals), and Seldja). A foraminiferal packstone is present near the top of the latest Maastrichtian Zone CF1 at El Kef and Elles II and marks a £ooding surface, winnowing and condensed sedimentation.

(goethite) up to 1 cm in diameter are also present. An important sedimentological change occurs Many intervals are mottled due to ichnofaunal at the top of the sequence. In the 30 cm thick activity and individual traces are rarely preserved interval directly underlying the K^T boundary, due to the soft pelitic sediment. In several hori- gray marls ¢rst grade into gray calcareous silt- zons, however, we determined Chondrites, Plano- stones and then into gray calcarenitic marly lime- lites and possibly Teichichnus, whereas megafos- stones, both of which form layers of 5 and 8 cm sils appear to be absent. thick respectively. Overlying this interval is a 5^7 8

Fig. 4. Bulk rock compositions at (A) El Melah, (B) El Kef, (C) Elles and (D) Seldja. Note that the dominant sedimentary components are phyllosilicates and calcite with temporally restricted in£ux of quartz, plagioclase and K-feldspar. A major composi- tional change occurs at the K^T boundary in all sections examined, though no exotic minerals were observed at this interval. cm thick yellow calcarenitic marly limestone con- foraminiferal sand and none with the green clay sisting primarily of planktic foraminiferal tests that overlies the calcarenitic layer. The upper sur- (foraminiferal packstone). This sediment is cross- face of the yellow calcarenitic marly limestone is bedded and burrowed (Fig. 3). Burrows are ap- undulating and likely represents an erosional dis- proximately 5 mm in diameter, unbranched, and conformity. reach a length of a few centimeters. Most are A plastic green clay, between 0.2 and 1 cm horizontal or oblique, but a few almost vertical thick, ¢lls the depressions of the calcarenite and shafts initiate in the upper surface of the bioclastic underlies a 2^4 mm thin rust-colored ferruginous layer. Even these tubes are packed with yellow layer interpreted as the K^T boundary red layer. 9

No burrows appear to be present in the green clay are less ¢ssile for several meters upsection. Bio- and none cross through the red layer. Overlying turbation ¢rst reappears at 50 cm above the red the red layer is a 1^2 cm thick plastic green clay layer. The stratigraphic interval considered in this followed by a second very thin layer of rust-col- paper covers the uppermost meter of the Maas- ored ferruginous material and a gypsum goethite trichtian (CF1), the K^T boundary (P0) and the crust (Fig. 3). Upsection, the clay grades into ¢rst meter of the Danian (P1a) (zonation after dark ¢ssile shales with small goethite concretions Keller et al., 2002). (Zone P0). The lowermost 10 cm of these shales Paleoceanographically, the Elles section was lo- are black, rich in organic matter, and contain rare cated closer to the emerged areas of the Kasserine casts of nuculanid bivalves. The ¢ssile black Island and hence received a higher in£ux of detri- shales grade into gray and light gray shales which tus than El Kef (Burollet, 1956). Paleodeposition 10

occurred in a middle to outer neritic environment The Ain Settara section was deposited in a (100^250 m). Due to the proximity to the Kasser- more proximal environment to the Kasserine Is- ine Island, sediment accumulation rates are signif- land than either Elles or El Kef and at inner to icantly higher than at El Kef with an average of 3 middle neritic paleodepths. The largely missing cm/ka at Elles II for Zone CF1, as compared with Zone P0 interval and very low sedimentation 2cm/ka for El Kef for the equivalent interval, and rate in Zone P1a (0.9 cm/ka) re£ect hiatuses at 2.3 cm/ka for the P0^P1a interval as compared the K^T boundary and at the top of Zone P0. with 1.9 cm/ka for El Kef (time scale based on Cande and Kent, 1995). 3.5. Seldja K^T

3.4. Ain Settara K^T Oued Seldja is located about 200 km from El Kef, in a gorge near Metlaoui (for directions see The Ain Settara section is located in the Ka- Keller et al., 1998). The outcrop spans the south laat-Senan area about 50 km southeast of El Kef £ank of a W^E stiking anticline and the beds dip and northwest of Elles (Fig. 1, see Dupuis et al., steeply (60^80³) to the south. The uppermost in press, for precise location). Only the top 60 cm Maastrichtian interval consists of 2.4 m of gray of the upper Maastrichtian gray silty marls were to brown silty shales, clays and clayey siltstones collected. At the top of this interval, and about 2 which are overlain by a 15 cm thick bed of yellow cm below the K^T red layer and dark clay is a silty sandstone rich in reworked ¢sh scales and weak disconformity marked by truncated Chon- teeth. The undulating contact between these lith- drites burrows (unbranched). A 2 cm thick dark ologies indicates erosion as also suggested by silty shale overlies the disconformity (Figs. 2, 3). truncated burrows and the presence of mud clasts The top of this silty shale is also marked by Chon- from the underlying sediments within the sand- drites burrows some of which are vertical, trun- stone layer (Figs. 2, 3, Keller et al., 1998). Above cated and in¢lled with the overlying dark clay. this layer are 1.15 m of gray silty shales and clays This suggests another disconformity centered at that contain early Danian (Zone Pla) planktic for- the K^T boundary. aminiferal assemblages. This interval is followed Similar to Elles I, the K^T transition is charac- by a 25 cm thick bed of gray sandy phosphate. terized by a 0.3^0.5 cm thick red layer sandwiched The lower contact of this bed is marked by an between two thin layers of secondary gypsum, undulose surface, manganese crusts and nodules goethite and jarosite. No burrowing is observed that suggest an extended period of erosion and across the red layer and in the overlying dark 2 non-deposition. The lower 5 cm of the phosphate cm thick boundary clay. Above the boundary clay layer contain mud clasts and phosphate-¢lled bur- layer is a 1.5 cm thick silty layer containing coars- rows extended into the underlying shales. The er grains of mainly diagenetic calcite (Fig. 3). This phosphate layer is rich in ¢sh remains (shark silt layer has been interpreted as a small storm teeth, ¢sh scales), small bivalves (mostly nucula- deposit by Dupuis et al. (in press). The presence nids) and gastropods. The ¢rst 5 cm above the of reworked transported species in the sample phosphate layer consist of gray-green marly silt- from this layer con¢rms the resedimented nature stone rich in microfossils (foraminifera and ostra- of this event bed which represents higher hydro- cods) with small (0.5^1 cm) clasts of phosphate. dynamic conditions at the P0/P1a boundary Upsection, marly siltstones (P1c) grade into (Luciana, 2002). It is likely that this event also brown siltstone with increasingly abundant terres- represents an erosional disconformity which could trial plant debris. be responsible,U at least in part, for the unusually During the K^T transition, the sediments were thin boundary clay (2 cm thick) in this section, as deposited in the shallow Gafsa Basin at inner ner- compared with Elles and El Kef. Upwards, the 70 itic depths (Fig. 1). To the south, the basin was cm sediments analyzed grade from dark gray ¢s- connected to the evaporite Sahara platform which sile shaley clay to gray (sometimes silty) shale. was separated from the Tethyan realm to the 11

north by the Kasserine Island. The interchange stone layers that spans through the P1d^P2 inter- with the open sea was therefore restricted and val. probably further hampered by small uplifted areas At El Kef and Elles, bulk rock compositions to the east and west which could have acted as di¡er from El Melah primarily in the continued barriers to circulation (Burollet, 1956; Burollet low calcite values through Zones P0 and Pla and Oudin, 1980; Sassi, 1974). Sediment deposi- (Fig. 4B, C). This di¡erence, however, may be tion occurred largely in restricted seas that £uctu- largely due to the more condensed section at El ated between inner neritic to coastal environ- Melah where P0 is only 10 cm thick and Pla is less ments. Tectonic activity and erosion of the than 1.5 m thick. The overall bulk composition Kasserine Island contributed to a constant though during the uppermost Maastrichtian Zone CF1 at variable terrigenous in£ux of sediments. No sedi- El Kef is similar to El Melah, with relatively high ment accumulation rate has been calculated due calcite (40^57%) and phyllosilicates (42^57%) and to the numerous hiatuses detected in this section. minor quartz (2^4%, Fig. 4B). Though in the top 30 cm of Zone CF1, calcite increases signi¢cantly (from 40 up to 57%) and then decreases (46%) 5 4. Bulk rock mineralogy cm below the K^T boundary clay layer. This calcite peak coincides with higher quartz (4%) and The dominant bulk rock composition of the K^ lower phyllosilicates (35^40%) and corresponds to T transitions in Tunisia is calcite and phyllosili- a similar interval at Elles II. At the K^T bound- cates, with minor amounts of quartz, feldspar ary, calcite drops to 5% at El Kef with phyllosi- (plagioclase and K-feldspar) and phosphate (F- licates (37%) and gypsum (37%) abundant. Goe- Ca apatite). Ankerite (Fe-rich dolomite) and pyr- thite is present just below and above the clay layer ite are sometimes present. Gypsum and hydroxide (1^2%). In the expanded Zone P0 at El Kef (and minerals such as goethite and jarosite are gener- Elles), calcite remains low (2^3%) and phyllosili- ally restricted to the K^T boundary clay. These cates reach maximum values (96%). The ¢rst sig- minerals are of secondary origin and re£ect late ni¢cant increase in calcite (up to 10%) coincides diagenetic processes. with decreased phyllosilicates at 47 cm above the At El Melah, the K^T transition is marked by a clay layer. At the P0^P1a transition, calcite de- sharp change in the bulk rock composition creases (3%) and phyllosilicates and quartz in- (Fig. 4A). Zone CF1 below the K^T boundary crease. Phyllosilicates (87^90%) reach maximum contains signi¢cant calcite (40^50%) and phyllosi- abundance in the lower part of Zone P1a, fol- licates (50^60%) with minor quartz (5^7%) and lowed by calcite (3^9%) and quartz (3^4%). Cal- plagioclase (1%). At the K^T boundary, calcite cite increases signi¢cantly (30^40%) 3 m above the and quartz drop abruptly to 1%, whereas phyllo- K^T clay layer up to the middle part of P1b, silicates and gypsum increase to 80 and 2%, re- whereas quartz and phyllosilicates are less abun- spectively. In the upper part of Zone P0, calcite dant with 2 and 50^60% respectively. Small increases to 20% and quartz to 6^7%, but both amounts of ankerite ( s 2%) coincide with these decrease at the P0^P1a transition. A short-term calcite maxima. The upper part of P1b and the increase in calcite (46%) in the ¢rst 10 cm of base of P1c Zones are again marked by decreasing Zone P1a parallels a decrease in phyllosilicates. calcite (18^25%) and coeval increasing detrital in- Upsection between Pla and the lower part of put (Fig. 4B). Plc, phyllosilicates decrease and calcite gradually At Elles I and II the K^T transition is marked increases with a temporary peak in Zone P1b (55^ by a sharp change in bulk rock composition, sim- 60%). Higher quartz (8^10%) and plagioclase (1^ ilar to El Kef and El Melah (Fig. 4C). The upper- 2%), but lower calcite (20^30%), mark the upper most Maastrichtian Zone CF1 is characterized by part of Plc, except for peak abundance (85%) near relatively high calcite and phyllosilicate contents the top of the section which marks the onset of (40^60%) and minor quartz (4^7%); feldspars, alternating deposition of marl and marly lime- goethite, jarosite and gypsum are not present in 12

that part of the section. There is an increase in 5. K^T sea-level changes: bulk rock mineralogy calcite (from 39 to 45%) 8 cm below the K^T, coinciding with a decrease in phyllosilicates. The Sea-level changes are commonly recognized K^T boundary clay layer contains very little cal- from ¢eld-based observations of lithological char- cite (1^2%) and quartz (2%), but abundant phyl- acteristics and laboratory analysis of bulk rock losilicates (52%) and gypsum (45%). Plagioclase compositions. In open marine settings, the carbo- (2%) and hydroxide minerals, such as goethite nate/detritus (mainly phyllosilicates and quartz) (1%) and gypsum of secondary origin are present ratio is a useful index for sea-level £uctuations. and re£ect late diagenetic processes. At Elles, sim- Increased carbonate content re£ects generally a ilar to El Kef, calcite remains low ( 6 15%) more distant detrital source and thus deeper water through Zone P0 and the lower part of P1a and conditions, whereas an increase in detritus indi- detrital in£ux is minor (quartz 6 4%, K-feldspar cates a more proximal source and consequently 2%, plagioclase 6 3%, Fig. 4B, C). a lower sea-level or shallower water environment. At the shallow inner neritic Seldja section, cal- Hiatuses or disconformities are indicated by non- cite deposition is nearly half (20^25%) of that in deposition and erosion surfaces, including bur- deeper water sections to the north (Fig. 1), where- rowed or semi-lithi¢ed omission surfaces in deep- as phyllosilicate deposition is higher ( s 70%, er waters. Bored and encrusted hardgrounds with Fig. 4D). Above the disconformity which marks phosphate and/or glauconite indicate shallow the K^T boundary (Pla/CF1), there is a major water environments, non-deposition and less com- change in bulk rock composition to dominant monly £ooding surfaces (Donovan et al., 1988; quartz (30^50%) and phosphate, increased feld- Loutit et al., 1988; Robaszynski et al., 1998; Vin- spar and plagioclase (11 and 13% respectively) cent et al., 1998). Calcisiltite and calcarenitic and decreased phyllosilicates ( 6 50%). The over- marly limestones enriched in foraminifera, such lying 25 cm thick silty phosphate layer (Zone Pla) as those observed near the top of the Maastrich- contains 16% calcite, 60% phyllosilicates, low tian at El Kef and Elles, represent current win- quartz and feldspar, and the ¢rst occurrence of nowing which is commonly associated with trans- phosphate (F-Ca apatite). A 30 cm thick phos- gressive system tracts. In addition to these major phate layer (V37% phosphate) marks a discon- sedimentary criteria, variations in the kaolinite/ formity between Zones Pla^Plc. Near the top of smectite (K/SM) ratio can also be used to infer the phosphate layer gypsum of late diagenetic ori- the proximity of the source area. Since kaolinite is gin is abundant (28^38%), whereas phyllosilicates more abundant in coastal areas and smectite in (60%) and calcite (40%) dominate above this in- open marine environments, the K/SM ratio may terval. also re£ect sea-level changes, in addition to cli- Bulk rock and clay mineral compositions have matic variations (Adatte and Rumley, 1989; not been measured by us for the Ain Settara sec- Chamley et al., 1990). tion. These data have been processed by Dupuis Recognition of sea-level £uctuations in our sec- et al. (in press). tions is thus based on lithological characteristics (e.g. hardgrounds, erosional surfaces, particle size variations) and the ratio of calcite/detritus (where

Fig. 5. Sea-level changes across the K^T transition at El Melah, El Kef, Elles and Seldja based on lithological variations, biostratigraphic data and bulk rock compositions (calcite/detritus ratios). In open marine settings, the calcite/detritus ratio (mainly phyllosilicates and quartz) is a useful index for sea-level £uctuations. Increased carbonate content re£ects a more distant detrital source and thus deeper water conditions, whereas an increase in detritus indicates a more proximal source and consequently a lower sea-level or shallower water environment. Note that the K^T boundary corresponds to a transgressive interval. The calcite/ detritus ratio indicates sea-level lowstands in Zones CF1, near the P0^Pla and P1b^P1c transitions. Note that two hiatuses are present at Seldja, at the K^T boundary (Zones P0, the lower part of Pla and upper part of CF1 are missing) and in the Danian where Zone Plb and upper part of Pla are missing. 13

CORRECTED PROOF 14

detritus includes quartz, phyllosilicates, plagio- T boundary reached a maximum in Zone P0 as clase and K-feldspar). Increased abundance of de- suggested by condensed clay deposition world- tritus may re£ect increased erosion and lower sea- wide (Brinkhuis and Zachariasse, 1988; Keller, levels, whereas increased calcite and decreased de- 1988a,b; Donovan et al., 1988; Baum and Vail, tritus suggest decreased erosion during higher sea- 1988; Keller and Stinnesbeck, 1996; Stinnesbeck levels. In addition, sea-level changes di¡erentially et al., 1996). The very low carbonate content in in£uence sediment deposition at various paleo- Zone P0 is likely related to the mass extinction of depths with sediment starvation occurring during tropical and subtropical planktic foraminifera and sea-level rises and sediment erosion during falls, calcareous nannofossils (Keller et al., 1995; Luci- particularly in shallow neritic areas. The Tunisian ana, 2002; Pospichal, 1994). The upper part of P0 sections analyzed span sedimentary environments is missing in the shallower sections of Ain Settara from inner neritic to middle and outer neritic and where P0 is only 3 cm thick, and at Seldja where upper slope settings (Fig. 1) and provide strong P0 is missing (Fig. 3). These hiatuses mark a sea- evidence for sea-level £uctuations across the K^T level lowstand and erosion in shallow water sec- boundary. tions at the P0^Pla boundary. A short hiatus at The calcite/detritus ratio in the Tunisian sec- the P0^Pla boundary has been observed in sections (Fig. 5) correlates with sea-level changes in- tions worldwide (MacLeod and Keller, 1991a,b; ferred from lithological and biostratigraphic data. Schmitz et al., 1992; Keller et al., 1993; Keller, At El Kef and Elles, this ratio averages 1.5 and 2 1993). respectively during the upper Maastrichtian A rising sea-level is indicated in Zone Pla (upper CF1) and indicates a relatively high sea- marked by increasing calcite deposition and in- level with signi¢cant terrigenous in£ux from the creasing calcite/detritus ratios which reach a max- nearby Kasserine Island (Fig. 1). In both sections imum in Zone Plb (Fig. 5). A sea-level lowstand the calcite/detritus ratio decreases to 1.0 between and short hiatus is associated with the Pla^Plb 50^100 cm below the K^T boundary and suggests boundary in many sections (Macleod and Keller, a lower sea-level accompanied by increased detri- 1991a,b; Keller and Benjamini, 1991; Canudo et tal in£ux (Fig. 5). Within the top 30^40 cm of the al., 1991; Keller and Stinnesbeck, 1996), though Maastrichtian the calcite/detritus ratio increased not observed in Tunisia, except at the shallow to about 2.0 and re£ects a signi¢cant sea-level neritic Seldja section. At Seldja, Plb and much rise which correlates with the foraminiferal pack- of Pla is missing due to a hiatus (Fig. 5). At the stone at El Kef and Elles (Fig. 3). Increased smec- Plb^Plc boundary the calcite/detritus ratio de- tite in this interval (Fig. 6) suggests that this sea- creases to 6 0.5 at El Kef and El Melah indicat- level rise was accompanied by drier climatic con- ing increased terrigenous in£ux and a lower sea- ditions (Chamley, 1989). The foraminiferal pack- level. A hiatus is widespread at the Plb^Plc stone likely accumulated by current winnowing boundary (MacLeod and Keller, 1991b), includ- and suggests the onset of a period of starvation ing at Seldja and re£ects a major sea-level regres- in the basin. At Elles I and II the packstone is sion. marked by an undulating lower surface that suggests erosion. An erosion surface is also present just below the K^T boundary at Ain Settara 6. K^T climate changes: clay mineralogy (Fig. 3). At El Melah, the foraminiferal packstone is not observed and its absence is likely due to the Clay mineral assemblages re£ect continental deeper water setting at this location. At Seldja, morphology and tectonic activity, as well as cli- this interval may be missing due to a hiatus that mate evolution and associated sea-level £uctua- spans the K^T boundary, though increasing cal- tions (Chamley, 1989, 1997; Weaver, 1989; Li et cite/detritus values near the top of the Maastrich- al., 2000). Illite and chlorite are considered com- tian also suggest a rising sea-level (Fig. 5). mon byproducts of weathering reactions with low The sea-level rise that began just below the K^ hydrolysis typical of cool to temperate and/or dry 15

Fig. 6. Clay mineral compositions ( s 2 Wm clay fraction, relative abundance) across the K^T transition at El Kef, Elles and Seld- ja. Note the peaks in smectite just below and at the K^T boundary and in the lower part of Zone P1a at El Kef and Elles. Smectite content is lowest just above the K^T boundary and Zone P0 and gradually increases in Zone Pla. 16

climates. Kaolinite is generally a byproduct of linite in all sections. Mica is generally a minor highly hydrolytic weathering reactions in perenni- component. Similar smectite and kaolinite aver- ally warm humid climates and its formation re- ages persist in the K^T boundary clay layer, quires a minimum of 15³C (Gaucher, 1981). The though smectite is generally high (V38% at El- presence of abundant smectite is generally linked les). In Zone P0 smectite signi¢cantly decreases to transgressive seas and warm climate with alter- (to V10^15%) at El Kef and Elles (P0 is missing nating humid and arid seasons, but can also re- at Seldja) and kaolinite increases (up to 85%), £ects volcanic activity (Chamley, 1989, 1997; De- whereas chlorite remains low (3^7%). About 40 coninck, 1992). The kaolinite/smectite ratio is a and 30 cm above the P0^P1a transition smectite climate proxy that re£ects humid/warm to more brie£y peaks at 51% at El Kef and at 40% at Elles dry and seasonal climate variations (e.g. Robert respectively (Fig. 6). Thereafter, smectite de- and Chamley, 1991; Robert and Kennett, 1992). creases to 9% at El Kef and then gradually in- The ratio of kaolinite to smectite (K/SM) can creases to 40% with peak abundance in chlorite therefore be used as proxy for climate change. and minimum values (47%) in kaolinite. At Elles, smectite gradually increases in the lower part of 6.1. Clay mineral results Pla and reaches maximum values of 38% similar to El Kef. At Seldja, most of Zone Pla is missing Kaolinite is the most abundant clay mineral and the interval present averages 20% smectite (mean value of 67%) in the K^T transitions of and 70% kaolinite. At El Kef, the P1a^P1b tran- Tunisia. Smectite, chlorite and mica are minor sition is characterized by decreased smectite (7%) components and average 18, 11 and 4%, respec- and increased kaolinite (80%). Smectite increases tively. All identi¢ed clay minerals are known from upsection to an average of 29% in Zone P1b and normal deposition or pedogenic environments and kaolinite decreases to an average of 55%. Similar exist in various amounts within Cretaceous and values are observed in Zone Plc at Seldja (Fig. 6). Paleogene sediments. Smectite is absent at El Mel- ah probably due to post-depositional burial linked 6.2. Inferred climate changes with increased tectonic activity (Tell thrusting, Burollet, 1956). Relative abundances and clay During the latest Maastrichtian to early Dan- mineral ratios of the El Melah section are there- ian, the Tunisian sections were located on a con- fore not comparable with other sections and tinental platform which experienced little hydro- hence not included in the following discussion. dynamic activity and hence little mineral At El Kef, Elles and Seldja the uppermost segregation that could mask or exaggerate the cli- Maastrichtian (CF1 Zone) sediments are domi- mate signal (Adatte and Rumley, 1989; Chamley, nated by kaolinite ( s 65%) and variable (10^ 1989; Monaco et al., 1982). Since kaolinite is usu- 38%) but gradually increasing smectite contents ally more abundant in coastal areas and smectite (Fig. 6). Chlorite is variable ( 6 20%), but reaches in open marine environments, the K/SM ratios maximum values (30%) just below the K^T may also re£ect sea-level changes. But primarily, boundary at El Kef and decreases upsection. variations in K/SM ratios are linked to climate The variability in smectite and chlorite is gener- changes. The smectite from the K^T layer is iden- ally compensated by decreasing or increasing kao- tical to the smectite within uppermost Maastrich-

Fig. 7. Kaolinite/smectite ratios across the K^T transition at El Kef, Elles and Seldja and inferred climatic evolution. The high K/SM ratios, the low content in chlorite with mica nearly absent indicate an overall warm and humid climate from the upper Maastrichtian Zone CF1 to the lower Paleocene Zone P1c. Note the four maxima in kaolinite/smectite ratio observed in the topmost Maastrichtian, in the upper part of P0, in the lower part of P1a and at the P1a/P1b transition (El Kef), re£ecting more humid conditions. A sea-level curve based on various sedimentological, biostratigraphical and geochemical proxies is superimposed over the K/SM ratio for El Kef. Decreases in K/SM ratios correspond to generally increased smectite during transgressive intervals under more seasonal conditions. 17 18

tian and earliest Danian sediments. ESEM and tions on land during the ¢rst 30^40 ka of the EDX analyses reveal a Fe-Al enriched and not Danian. Peak K/SM ratios are also observed in well crystallized smectite which typically develops the lower part of Zone Pla, about 1.5 m above the in soils under intermediate climatic conditions K^T boundary at Elles and El Kef, and suggest with seasonal contrasts (Chamley, 1989, 1997). an interval of warmer humid climate (Fig. 8a). At At Elles and El Kef, the high K/SM ratios, the Seldja, the peak K/SM ratio in Zone Pla at 80 cm low content in chlorite with mica nearly absent above the K^T boundary may correlate this warm indicate an overall climate trend characterized humid period (upper part of Pla is missing at by humid and warm conditions across the K^T Seldja). Lower K/SM ratios during most of transition (Zones CF1 to P1c, Figs. 7 and 11). Zone Pla at Elles and El Kef suggest a drier cli- Within this interval are four K/SM maxima lo- mate with increasing seasonality. At El Kef, a cated respectively 80 cm below the KTB, in the maxima in the K/SM ratio at the Pla^Plb bound- lower part of P0, in the lower part of P1a and at ary marks increased precipitation and is associ- the P1a^P1b transition (Figs. 7, 8A). The second ated with a lower sea-level and high detrital in- K/SM maximum, located in the lower part of P0 £ux. At Seldja, a maximum in kaolinite is just above the red clay layer, has also been ob- observed in the phosphatic sandstone of Zone served at Caravaca (Kaiho et al., 1999). Within P1c (Fig. 6) which re£ects a condensed interval this main trend, relatively low K/SM ratios are during a rising sea-level. These data suggest that located at 15^20 cm below the K^T red layer, the climate across the K^T transition (Zones CF1 near the P0^P1a transition, in the middle of P1a to Plc) in Tunisia alternated between warm and and in P1b. These K/SM minima indicate, within humid periods and drier periods with increasing the overall humid condition trend, periods of in- seasonality. In general, warm/humid (perennially creasing seasonality and possibly drier climatic wet) periods coincide with low sea-levels, whereas conditions. These relatively low K/SM ratios cor- increased seasonality and drier climates coincide relate with high sea-levels observed in sections with higher sea-levels inferred from bulk rock globally (MacLeod and Keller, 1991a,b; Keller compositions (Fig. 8B). and Stinnesbeck, 1996) and also inferred based on the calcite/detritus ratio in this study. At El Kef and Elles, relatively low K/SM ratios 7. Organic matter in the uppermost part of CF1 suggest increased seasonal alternations in the rainfall regime (ST, Total organic carbon (TOC) in marine sedi- Fig. 7). The subsequent sea-level transgression ments is generally a geochemical proxy for pri- marked by the foraminiferal packstone (top 40 mary productivity and carbon burial linked to cm of Zone CF1 at El Kef and top 25 cm at climate, erosion and sea-level £uctuations. During Elles), is associated with increased kaolinite fol- low sea-levels, or perennially wet climates, TOC lowed by a decrease 15^20 cm below the KT red values are generally high as a result of enhanced layer (Fig. 6). The kaolinite decrease (to 50%) terrestrial organic matter in£ux associated with coincides with a strong increase in smectite (to increased erosion and carbon burial. During 40%) which persists into the K^T clay layer at high sea-levels, TOC values are generally lower, El Kef and Elles. At both El Kef and Elles sec- except in special circumstances, such as the K^T tions, the K/SM ratios in Zone P0 (Fig. 7) suggest boundary clay where organic carbon is concen- a series of alternating warm humid (WH) and trated at a £ooding surface associated with de- seasonally wet (ST) climate £uctuations beginning creased productivity or anoxia. at the K^T boundary with the ¢rst 2^10 cm warm Average values of total organic carbon (TOC) and humid, between 10^20 cm seasonally temper- are relatively high at El Kef and Elles (0.45^ ate, 20^35 cm warm humid, 35^70 cm seasonally 0.50%), but relatively lower at El Melah (0.30%) temperate (Fig. 7). These clay mineral variations and Seldja (0.20%, Fig. 9). In the uppermost 20^ re£ect alternating drier and more humid condi- 30 cm of the Maastrichtian (CF1), TOC values 19

Fig. 8. (A) Short-term trends in the kaolinite/smectite (K/ SM) ratio at El Kef an Elles, suggesting a series of alternating perennially wet (WH) and seasonal temperate (ST) climate at the KT transition. (B) Summary of hiatuses based on biostratigraphic data and sedimentological observations (e.g. disconformities, truncated burrows, lithological changes), sea-level changes inferred from bulk rock compositions, lithological and sedimentological characteristics and biostratigraphy, and climate changes inferred from clay mineral data across the K^T transition in Tunisia. Note that short hiatuses are identi¢ed at the top of Zone CF1 (Seldja, Ain Settara, Elles), at the P0/P1a, Pla^Plb and P1b^P1c boundaries, all of which correlate with low sea-levels and have been recognized globally (MacLeod and Keller, 1991a,b; Keller and Stinnesbeck, 1996). The P1a^P1b hiatus is not well developed in Tunisia. High sea-levels generally correspond to increasing seasonality and drier conditions, whereas low sea-levels correspond to increasing humidity and possibly warmer conditions.

increase and reach maximum values in the K^T boundary red clay layer in all sections, except Seldja where this interval is missing. The red clay layer is therefore marked in all sections by maximum TOC values (1.5% at El Kef, 0.65% at El Melah, 0.8% at Elles). At Elles and El Kef, TOC values decrease in Zone P0 to an average of 0.45% and at El Melah to 0.20%. Upsection in Zones Pla^Plb, TOC contents £uctuate between 0.15 and 0.7% at El Kef and 0.1 and 0.35% at El Melah. At Elles and El Kef, the relatively high average TOC values recorded in the P0 Zone can be explained by sediment starvation during sea-level rise. At Seldja, organic matter is lower due to erosion and oxidation. At El Kef and Elles, the overall TOC trend indicates the continuous presence of signi¢cant organic matter with a maximum coinciding with the K^T boundary. These relatively high TOC values agree well with the perennially wet and warm climate inferred from clay minerals and the associated accelerated run- o¡ from continents which prevailed during the uppermost Maastrichtian. TOC contents increase at El Melah, El Kef and Elles about 30^40 cm below the K^T boundary, coincident with the onset of the transgressive interval which culminates in the K^T clay layer (P0). Maximum TOC content in P0 may be of multiple origins, including 20 21

Fig. 9. Total organic carbon content (TOC) across the K^T transition at El Melah, El Kef, Elles and Seldja. Note that TOC content begins to increase just before the K^T boundary and reaches maximum values at the K^T boundary layer. Peak TOC values at the Pla^Plc hiatus interval at Seldja mark a phosphate-rich layer.

Fig. 10. Late Cretaceous to early Tertiary bulk rock compositions in a composite section based on Elles (upper Campanian to lower Maastrichtian Zone CF7) and El Kef (lower Maastrichtian Zone CF7 to Paleocene Zone Plc). Stippled and numbered intervals mark sea-level lowstands 1^8 identi¢ed based on biostratigraphic, sedimentological, mineralogical and geochemical proxies (Li et al., 1999, 2000; this study). Note that major peaks in phyllosilicates and quartz mark major sea-level lowstands. The long- term increase in phyllosilicates beginning in Zone CF5 culminates in the lower Danian Zones P0 and Pla and appears to be related to increased humidity, weathering and runo¡ from nearby terrestrial areas (e.g. Kasserine Island) and decreased productivity in the early Danian coupled with the extinction of the large tropical and subtropical planktic foraminiferal species near the K^T boundary. 22

biogenic material derived from surface waters of Zone P0 and the lower part of the Zone P1a. the ocean (Lindinger, 1988) and organic debris During the same time, calcite decreases and derived from adjacent continental regions, as ob- quartz in£ux remains stable between 3 and 6%. served in the upper Maastrichtian (Li et al., 2000). This long-term trend in bulk rock compositions Rock-Eval analyses indicate that the organic mat- indicates a general increase in detrital in£ux be- ter is primarily of terrigenous origin (Type III, Li ginning in the late Campanian and increasing et al., 2000). The TOC enrichment is therefore through the Maastrichtian into the early Danian. likely due to concentration and/or preservation This trend re£ects a major regional change in of continental organic matter during condensed weathering processes and hence climate. The over- sedimentation associated with a high sea-level all increase in detrital in£ux correlates with the (maximum £ooding surface). Large-scale defolia- global cooling trend observed in oxygen isotope tion of terrestrial plants due to an impact (Orth et records from Sites 525 and 690 (Barrera, 1994; Li al., 1981; Tschudy and Tschudy, 1984; Saito et and Keller, 1998a, b). al., 1986; Lindinger, 1988) and wild¢res (Wolbach Phyllosilicate/calcite £uctuations within this et al., 1985) are unlikely in this region since no long-term trend also re£ect sea-level changes `fern spike' or other major change is observed in with high detrital abundances generally associated palyno£ora (Meon, 1990; Meon et al., 2002). with low sea-levels and high terrestrial in£ux (no- tably K-feldspar and quartz). Based on these proxies, seven major sea-level lowstands can be 8. Campanian^Maastrichtian trends recognized between the late Campanian and Maastrichtian (stippled pattern, Fig. 10). An ad- Lower Maastrichtian and Campanian outcrop ditional sea-level regression has been recognized exposures are discontinuous at El Kef. For this near the top of Zone CF1 in this study as noted reason this interval (CF10^CF7) was sampled at earlier (Fig. 5C). The seven sea-level £uctuations Elles and the lower Maastrichtian to Danian at El correlate well with those previously identi¢ed Kef (CF7^Plb). For lithological descriptions see based on various geochemical (stable isotopes, Li et al., 1999, 2000). Here we present previously Sr/Ca ratios, TOC) faunal (macro- and microfos- unpublished bulk rock and clay mineral composi- sils, trace fossils) and sedimentological (marl/lime- tions through this interval. Sediment accumula- stone transitions, hardgrounds, hiatuses) proxies tion rate averages are very low (0.5 cm/ka) for (Li et al., 1999, 2000). Low sea-levels are generally the upper Campanian^lower Maastrichtian inter- associated with increased terrigenous in£ux, low val as a result of hiatuses (see Li et al., 1999). For kaolinite/chlorite+mica ratios, high TOC and high the upper Maastrichtian sediment accumulation Sr/Ca ratios, whereas high sea-levels are generally rates average 2 cm/ka (Li and Keller, 1998a,b,c). associated with the reverse conditions. Bulk rock compositions at Elles and El Kef Clay mineral compositions also show short- indicate cyclic variations in calcite and detrital term and long-term trends. Short-term variations minerals as well as a long-term trend of increasing are seen in cyclic patterns of kaolinite, chlorite phyllosilicates and decreasing calcite (Fig. 10). and mica which suggest alternating humid/warm For example, the long-term trend in phyllosili- and seasonal/dry climates (Fig. 11). Long-term cates from 6 10% in the late Campanian Zone trends are re£ected in the kaolinite and smectite CF10 reaches maximum values of 80^90% in abundances and represent the long-term cooling

Fig. 11. Late Cretaceous to early Tertiary clay mineral compositions in a composite section based on Elles (upper Campanian to lower Maastrichtian Zone CF7) and El Kef (lower Maastrichtian Zone CF7 to Paleocene Zone Plc). Stippled intervals mark sea- level lowstands. Note the absence of kaolinite in the upper Campanian Zone CF10, rapid increase in kaolinite in Zone CF9 and again in CF4, and near the K^T boundary. These increases in kaolinite mark a climate trend from arid/dry in the late Campa- nian to increasingly perennially wet and warm during the Maestrichtian^lower Paleocene interval, in the southwestern Tethys region. 23 24 25

trend with increased weathering. During the late 9. Other sea-level and climate proxies Campanian (CF9^CF10), smectite dominates and kaolinite is nearly absent which suggests a season- Strontium concentrations in marine sediments al to arid climate in this part of northern Africa. have been successfully used to infer sea-level £uc- From CF8 upwards, kaolinite dominates with tuations (Graham et al., 1982; Renard, 1986; maximum abundance between the latest Maas- Stoll and Schrag, 1996). Sr/Ca ratios measured trichtian (CF2^CF1) and early Danian (P1a). in well-preserved planktic foraminifera from the Smectite, chlorite and mica are present in variable Elles and El Kef sections yield relatively accurate abundances. This clay mineral pattern suggests records of Sr £uctuations in seawater and conse- that from the latest Campanian Zone CF8 quently sea-level changes (Li et al., 2000). A through the Maastrichtian, Tunisia was character- strong correlation exists between variations of ized by humid (high precipitation) and warm cli- Sr/Ca ratios in planktic foraminifera with global matic conditions, though with increased season- sea-level £uctuations due to the shifts in sink and ality in the late early Maastrichtian (Zones source of Sr during sea-level falls and rises. All CF6^CF5) as suggested by the kaolinite/chlori- major increases in Sr/Ca values coincide with sea- te+smectite ratio (Fig. 11). level lowstands (Fig. 12). But from CF5 upwards, The brief smectite maximum near the base of the Sr/Ca record also shows an overall increase in Pla at El Kef and Elles (Fig. 6) is anomalous in the Sr/Ca ratio which may re£ect a long-term this trend and may be explained by sea-level £uc- change in the Sr due to the increased weathering tuations. Smectite consists of ¢ne particles which rate indicated by bulk and clay mineralogy. are easily carried into slope and basin environ- At El Kef and Elles, TOC values are generally ments during transgressive seas, as was the case high during low sea-levels either as a result of in the earliest Zone Pla. Coarser clay particles, enhanced primary productivity, or more likely such as kaolinite, chlorite and mica, are trapped high terrestrial organic in£ux, as con¢rmed by in neritic environments (Adatte, 1987; Chamley, Rock-Eval analyses (Li et al., 2000). From CF6^ 1989; Weaver, 1989). During the late Maastrich- CF5 upwards, TOC values gradually increase tian (Zone CF4 upwards), the general clay miner- with maxima occurring during sea-level lows. al trend of increasing kaolinite appears to be This overall TOC increase re£ects increasing run- linked to a regional climate change towards o¡ due to intensi¢ed weathering on land under more humid, wet, though not necessarily warmer, perennially wet climatic conditions (Fig. 12). A climatic conditions with minor sea-level £uctua- reverse trend is observed in the K^T clay layer tions. Middle and high latitude stable isotope re- where high TOC contents is associated with a cords indicate that climate cooled signi¢cantly sea-level rise. during the CF4^CF3 time interval (Barrera, Sr87/Sr86 ratios are linked to weathering and 1994; Li and Keller, 1998a,b). Clay mineral data erosion and indirectly to sea-level and climate from Tunisia thus suggest that the late Maastrich- changes. At DSDP Site 525 in the South Atlantic, tian was a time of increased humidity in the low Sr87/Sr86 ratios show a continuous rise throughout latitude Tethys, coinciding with the cooling ob- the Maastrichtian with maximum values between served in high latitudes. 300 and 400 ka (CF2^CF1 transition) before the K^T boundary (Fig. 12) as also observed in other studies (Sugarman et al., 1995; Hess et al., 1986;

Fig. 12. Summary of late Cretaceous sea-level and climate £uctuations inferred from mineralogical proxies (phyllosilicates, calcite, kaolinite) and various geochemical proxies, including Sr/Ca ratios, Sr87/Sr86 ratios, TOC, and magnetic susceptibility (Hansen et al., 1996). Stippled interval marks low sea-levels. Note that from the early late Maastrichtian onwards, the gradual increase in each of the proxies (except calcite) indicates a change in weathering rates due to increased humidity. Within this long-term change, the major sea-level regressions generally correlate with high Sr/Ca ratios, high TOC and high terrigenous in£ux. 26

Arthur et al., 1992; Frank and Arthur, 1999). atmospheric PCO2. Within this long-term climate Data from El Kef and Bidart (Vonhof and Smit, change, maximum humid conditions are reached 1997) indicate a further increase in Sr87/Sr86 ratios just before the K^T boundary (top CF2 and CF1 with a maximum at the P0^P1a transition Zones) and in the earliest Danian (P0 to basal (Fig. 12). A rise in Sr87/Sr86 ratios re£ects in- P1a, Figs. 8B, 11). Eight sea-level lowstands are creased weathering due to enhanced rainfall. In- identi¢ed between the late Campanian and the K^ tensive leaching of soils and sialic surface rocks T boundary with estimated ages and durations for located on the African craton likely contributed the ¢rst seven given by Li et al. (1999, 2000) as: to the Sr runo¡ to the ocean. Within this long- late Campanian V74.2 Ma, 73.4^72.5 Ma and term trend, Deccan Trap volcanism could have 72.2^71.7 Ma, early Maastrichtian 70.7^70.3 induced the 0.0004 decrease in Sr87/Sr86 ratio ob- Ma, 69.6^69.3 Ma and 68.9^68.3 Ma, and late served in the middle part of Zone CF1 at El Kef Maastrichtian 65.45 Ma. This study con¢rms (Vonhof and Smit, 1997; Martin and McDougall, these sea-level £uctuations based on bulk rock 1991). and clay mineral compositions and adds one addi- Magnetic susceptibility of sediments is linked to tional brief sea-level lowstand in the latest Maas- abundance of magnetic minerals and phyllosili- trichtian V25^100 ka before the K^T boundary. cates (Hansen et al., 1996; Urrutia-Fucugauchi, Major eustatic sea-level falls identi¢ed by Haq et 1997) and indirectly to enhanced continental run- al. (1987) coincide with Tethyan sea-level falls o¡ due to rainfall. Hansen et al. (1996) observed identi¢ed at 68.9^68.3 Ma, 70.7^70.3 Ma and that magnetic susceptibility for the latest Maas- V74.2 Ma. This suggests that these £uctuations trichtian at El Kef reached maximum values in are of eustatic origin. The late Maastrichtian sea- the upper part of Zone CF1 (Fig. 12) as also level low at 65.45 Ma is also likely of eustatic observed at Elles by Hambach (personal commu- origin as indicated by its presence in sedimento- nication, 1999). This magnetic susceptibility max- logical sequences worldwide (Keller and Stinnes- imum correlates with the sea-level lowstand iden- beck, 1996). A sea-level transgression marks the ti¢ed in this study in the upper Zone CF1 end of the Maastrichtian and maximum £ooding (Fig. 5A). coincides with the K^T boundary clay layer (e.g. Donovan et al., 1988; Baum and Vail, 1988; Kel- ler and Stinnesbeck, 1996; Stinnesbeck et al., 10. Discussion 1998). The early Danian sea-level falls near the P0^Pla and Pla^Plb boundaries have been identi- Sea-level and climate proxies, such as bulk rock ¢ed worldwide (MacLeod and Keller, 199la,b; and clay mineral compositions, TOC Sr/Ca and Keller and Stinnesbeck, 1996). Sr87/Sr86 ratios, indicate that Maastrichtian sedi- Stable isotope records from South Atlantic ments from the southwest Tethys initiate a grad- Sites 525 and 690C (Fig. 13, Barrera, 1994; Bar- ual climatic change towards more humid, though rera et al., 1997; Li and Keller, 1998a) indicate not necessarily warmer, conditions. This change that the cooling trend in sea surface waters began in weathering £ux, including the increasing CO2 some time in the late Campanian (CF10) and consumption, signi¢cantly contributed to climate reached maximum lows in Zones CF4^CF3 in cooling by e¡ecting an increased drawdown of the high latitude Site 690 and near the top of

Fig. 13. Summary of late Cretaceous to early Paleocene sea-level and climate £uctuations and the species richness in planktic foraminifera in Tunisia and correlation with the N18O record from the South Atlantic Site 525 (Li and Keller, 1998b). Note the early late Maastrichtian was a time of increased humidity in the southwestern Tethys and cooling in high latitudes. Species richness in planktic foraminifera correlates signi¢cantly with this gradual change in climate, weathering and nutrient import to the oceans. The pre-K^T boundary decrease in species richness in Zones CF1^CF2 (65.45 Ma) coincides with the onset of maximum humidity in Tunisia and a short warm event in the southern high latitudes. 27 28

CF3^CF2 in the middle latitude Site 525 (Fig. 13). begins in Zone CF2 and culminates at the K^T This cooling trend is interrupted by periodic boundary, correlates with maximum humidity warming in surface waters and the in£ux of (Fig. 13). The decline in species richness also cor- warmer deep waters in middle latitudes in Zones relates with increased terrestrial in£ux and conse- CF5 and CF6. A major short-term warming quently nutrients. Nutrient export to the ocean marks the interval between 200 and 400 ka before probably reached a maximum in the upper CF1 the K^T boundary (Li and Keller, 1998a, b). to lower Pla interval. Carbon isotopes indicate These major cooling and warming phases gener- that surface productivity decreased globally dur- ally coincide with sea-level £uctuations identi¢ed ing the early Danian and may have adversely af- in the southwestern Tethys (Figs. 10, 12). Some fected the watermass strati¢cation and nutrient authors suggest that the long-term Campanian^ balance which contributed to the low species di- Maastrichtian cooling may have been due to po- versity and slow recovery of marine plankton lar ice in Antarctica (Barrera, 1994; Abreu et al., after the mass extinction. 1998). If this hypothesis can be con¢rmed, then the late Cretaceous sea-level changes observed in Tunisia may be related to high latitude cooling 11. Conclusions and possible ice caps on Antarctica. Clay mineral data suggest that the high latitude (1) El Kef and Elles sections contain the most cooling during the late Campanian and Maas- expanded and complete sedimentary records trichtian was associated with increased humidity known to date and therefore are the appropriate in the southwestern Tethys. This re£ects climatic global standard stratotype and point (GSSP) and conditions that are similar to the last Quaternary additional GSSP sections respectively. glacial episode when the Sahara region experi- (2) Our observations based on biostratigraphi- enced humid conditions and greened with abun- cal, lithological and mineralogical data indicate dant vegetation. Maximum humidity character- that the K^T transition is marked by a sea-level ized the K^T boundary and may be linked to lowstand in the latest Maastrichtian about 25^100 greenhouse e¡ect induced by Deccan volcanism. ka below the K^T boundary. This sea-level fall is At this time, humidity was not restricted to the marked by increased detrital input at El Kef and Tethys, but reached well into the middle latitudes Elles and a short hiatus at Ain Settara. A rising as indicated by data from the South Atlantic Site sea-level marks the end of the Maastrichtian and 525 that suggests perennially wet conditions (sig- is expressed at Elles and El Kef by deposition of a ni¢cant increase in kaolinite) during the CF2^ foraminiferal packstone. A £ooding surface marks CF1 interval (Adatte et al., in preparation). the K^T boundary clay associated with condensed These changes in climate and associated oceanic sedimentation and maximum in£ux of terrestrial circulation appear to have signi¢cantly a¡ected organic matter. The P0^P1a transition is marked planktic foraminiferal species richness as well as by a sea-level lowstand corresponding to a short species richness in palyno£oras and most inverte- hiatus at Ain Settara where the upper part of P0 brates, except for inoceramids and rudistids which is missing and a period of non-deposition and went extinct. For example, species richness erosion marks the lower part of P1a. At Seldja, doubled for most of these groups in Zone CF6 P0 and possibly the topmost part of CF1 is miss- (as exempli¢ed in Fig. 13 for planktic foramini- ing. fera) following the ¢rst cooling maximum between (3) All the identi¢ed clay minerals are known 71 and 70 Ma and reached maximum richness in from normal deposition or pedogenic environ- CF4 and CF3 (Keller and Li, in press). Note that ments and exist in various amounts within the the maximum species richness coincides with in- Cretaceous and Paleogene. Thus there are no min- creasing calcite, high kaolinite and decreasing eralogical components that are unique or that phyllosilicates in Zones CF4^CF3 (Figs. 10, 11). could be related to an exotic event. Kaolinite in- The terminal decline in species richness, which creases from the late Maastrichtian into the early 29

Danian and indicates overall increased humidity. and subsequent mass extinction of tropical and Smectite peaks just below the K^T boundary and subtropical taxa coincides with maximum humid- near the base of Pla indicate a sea-level rise linked ity and terrestrial nutrient export to the oceans with drier climatic conditions. At both El Kef and which may have been associated with Deccan vol- Elles sections, the K/SM ratios in Zone P0 suggest canism and a bolide impact. a series of alternating warm humid and seasonal temperate climate (Fig. 7). Sea-level and climate £uctuations inferred from bulk rock and clay min- 12. Uncited references eral data indicate a consistent relationship of low sea-levels with high humidity and high sea-levels Adatte et al., 1998, 2000; Alvarez et al., 1980; with low humidity (increasing seasonality) in the Berggren et al., 1995; Courtillot et al., 1986; Hal- southwestern Tethys region (Fig. 8B). lam, 1992; Hansen and Toft, 2002; Mc Arthur, (4) Terrestrial organic matter, as well as Ir and 1994; McLean, 1985; Parsons and Takahashi, other trace elements, are concentrated at the K^T 1973; Said, 1978; Tschudy et al., 1984 boundary clay layer which globally coincides with a sea-level rise (maximum £ooding) and implies sediment starvation and a condensed interval Acknowledgements (Fig. 8B). These trace elements are therefore signi¢cantly elevated as a result of concentration and We thank Dr. M. Bel Haj Ali, Director of the additional in£ux due to extraterrestrial event must Tunisian Geological Survey, for hosting the 1998 be evaluated with respect to normal £ux. International Workshop on the K^T boundary in (5) Late Campanian to early Danian trends in Tunisia and for supporting the field excursion and bulk rock compositions indicate an increasing de- Dr. Habib Bensalem for arranging logistical trital in£ux that culminated during the K^T tran- support and guidance for the field excursion which sition and re£ect a global change in weathering made collection of samples from Ain Settara and and climate. At the same time, clay minerals in- Elles possible for participants. We thanks Jose dicate a long-term trend towards a more humid `Derrick' Richard for the sample preparation for climate which also culminated during the K^T XRD analysis, Philip Steimann for conducting transition. For the same interval, middle and TOC analysis, Jerry Baum and Herve¨ Chamley for high southern latitude oxygen isotope records in- their constructive reviews. This study was sup- dicate long-term climate cooling. This suggests ported by grants from NSF INT 95-04309, DFG that during the Maastrichtian humidity increased grant Sti 128/4-1 and the Swiss National Fund No. in low latitudes whereas high latitudes cooled. 8220-028367. The short-term warming just below the K^T boundary may be linked to Deccan volcanism and may have enhanced already humid conditions References in the Tethys region and increased continental runo¡. Abramovich, S., Keller, G., 2002. Planktic foraminiferal pop- (6) The species richness patterns of planktic ulation changes during the Late Maastrichtian at Elles, Tu- nisia. Palaeogeogr. Palaeoclimatol. Palaeoecol. foraminifera, invertebrates (except inoceramids Abreu, V.S., Hardenbol, J., Haddad, G., Baum, G.R., Drox- and rudistids) and palyno£ora correlate signi¢- ler, A.W., Vail, P.R., 1998. Sequence stratigraphy of the cantly with the observed changes in climate and European basins. Soc. Econ. Paleontol. Mineral. Spec. weathering. Doubling of species richness between Publ. 60, 75^80. Zones CF6 and CF5, following a major cool Adatte, T., Rumley, G., 1989. Sedimentology and mineralogy event, may be related to increased humidity and of Valanginian and Hauterivian in the stratotypic region (Jura mountains, Switzerland). In: Wiedmann, J. (Ed.), Cre- precipitation and the resultant increase in terrige- taceous of the Western Tethys Proceedings 3rd International nous in£ux, including terrestrial organic matter. Cretaceous Symposium. Scheizerbart'sche Verlagsbuchandl- The decrease in species richness in CF2^CF1 hung, Stuttgart, pp. 329^351. 30

Adatte, T., Stinnesbeck, W., Keller, G., 1996. Lithostrati- la Tunisie Centrale. Annales des Mines et de la Geologie, graphic and mineralogic correlations of near K^T boundary Tunis vol. 18, 388 pp. clastic sediments in northeastern Mexico: Implications for Burollet, P.F., Oudin, J.L., 1980. Paleöce©ne et Eoce©ne en Tu- origin and nature of deposition. Geol. Soc. Am. Spec. Pap. nisie, Pe¨troles et Phosphates. In: Geölogie Compareë des 307, 211^226. Gisements de Phosphates et des Pe¨troles. Bur. Rech. Geöl. Adatte, T., Keller, G., Li, L., Stinnesbeck, W., Zaghbib-Turki, Min. Doc. vol. 24, pp. 205^216. D., 1998. Climate and sea level £uctuations across the K-T Cande, S.C., Kent, D.V., 1995. Revised calibration of the geo- boundary in Tunisia: Warm and humid conditions linked to magnetic polarity timescale for the Late Cretaceous and the Deccan volcanism? In: International Workshop on Cre- Cenozoic. J. Geophys. Res. 100, 6093^6095. taceous-Tertiary Transition, O¤ce National des Mines, Di- Canudo, J.I., Keller, G., Molina, E., 1991. Cretaceous/Tertiary rection du Service Geologique, Tunis, Tunisie, 1998. Ab- extinction pattern and faunal turnover at Agost and Cara- stracts, pp. 7^10. vaca: S.E. Spain. Mar. Micropaleontol. 17, 319^341. Adatte, T., Bolle, M.P., De Kaenel, E., Gawenda, P., Winkler, Chamley, H., 1989. Clay Sedimentology. Springer-Verlag, Ber- W., Von Salis, K., 2000. Climatic evolution from Paleocene lin, 623 pp. to earliest Eocene inferred from clay-mineral: a transect Chamley, H., Deconinck, J.F., Millot, G., 1990. Sur l'abond- from northern Spain (Zumaya) to southern (Spain, Tunisia) ance des mine¨raux smectitiques dans les se¨diments marins and southern-east Tethys margins (Israel, Negev). GFF communs de¨pose¨s lors des pe¨riodes de haut niveau marin du Meet. Proc. Geol. Soc. Sweden 122, 7^8. Jurassique au Paleöge©ne. C. R. Acad. Sci. Paris 311, 1529^ Adatte, T., Keller, G., Li, L., in preparation. Short uppermost 1536. Maastrichtian greenhouse episode linked to Deccan, Evi- Chamley, H., 1997. Clay mineral sedimentation in the Ocean. dence from the high latitude South Atlantic (Site 525a). In: Paquet, H., Clauer, N. (Eds.), Soils and Sediments Min- Alvarez, L.W., Alvarez, W., Asaro, F., Michel, H.V., 1980. eralogy and Geochemistry. Springer, Berlin, pp. 269^302. Extraterrestrial causes for the Cretaceous-Tertiary extinc- Courtillot, V., Besse, J., Vandamme, D., Montignie, R., tion. Science 208, 1095^1108. Jaeger, J.J., Cappeta, H., 1986. Deccan £ood basalts at the Barrera, E., 1994. Global environmental changes preceding the Cretaceous/Tertiary boundary? Earth Planet. Sci. Lett. 80, Cretaceous-Tertiary boundary: Early-late Maastrichtian 361^374. transition. Geology 22, 877^880. Deconinck, J.F., 1992. Se¨dimentologie des argiles dans le Ju- Barrera, E., Huber, B.T., 1990. Evolution of Antarctic waters rassique-Cre¨tace¨ d'Europe occidentale et du Maroc. Me¨m. during the Maestrichtian:Foraminifera oxygen and carbon Habilit. Lille I, 266 pp. isotope Ratios, Leg 113. In: Barker, P.F., Kennett, J.P., et Donovan, A.D., Baum, G.R., Blechschmidt, G.L., Loutit, al. (Eds.), Proc. Ocean Drill. Prog. Sci. Res. 113, 813^827. T.S., P£um, C.E., Vail, P.R., 1988. Sequence stratigraphic Barrera, E., Savin, S.M., Thomas, E., Jones, C.E., 1997. Evi- setting of the Cretaceous-Tertiary boundary in central Ala- dence for thermohaline-circulation reversals controlled by bama. In: Wilgus, C.K., Posamentier, H., Ross, C.A., Ken- sea-level change in the latest Cretaceous. Geology 25, 715^ dall, C.G. (Eds.), Sea-Level Changes: An Integrated Ap- 718. proach. Soc. Econ. Paleontol. Mineral. Spec. Publ. vol. 42, Baum, G.R., Vail, P.R., 1988. Sequence stratigraphic concepts pp. 299^307. applied to Paleogene outcrops, Gulf and Atlantic basins. In: Dupuis, C., Steurbaut, E., Molina, E., Rauscher, R., Aschuler, Wilgus, C.K., Posamentier, H., Ross, C.A., Kendall, C.G. M., Tribovillard, N.P., Arenillas, I., Arz, J.A., Robaszynski, (Eds.), Sea-Level Changes: An Integrated Approach. Soc. F., Caron, M., Larque, P., (in press). Biotic evolution and Econ. Paleontol. Mineral. Spec. Publ. vol. 42, pp. 309^327. events across the Cretaceous/Tertiary (K-T) boundary in the Berggren, W.A., Kent, D.V., Swisher, C.C., Aubry, M.P., newly discovered Ain Settara section (Kalaat-Senan area, 1995. A revised Cenozoic geochronology and chronostratig- Central Tunisia). Palaeogeogr. Palaeoclimatol. Palaeoecol. raphy. SEPM 54, 129^213. Ferrero, J., 1965. Dosage des principaux mine¨raux des roches Brinkhuis, H., Zachariasse, W.J., 1988. Dino£agellate cysts, par di¡raction de Rayon X. Rapport C.F.P., Bordeaux, in- sea level changes and planktic foraminifera across the Creta- e¨dit. ceous/Tertiary boundary at El Haria, northeast Tunisia. Ferrero, J., 1966. Nouvelle me¨thode empirique pour le dosage Mar. Micropaleontol. 13, 153^190. des mine¨raux par di¡raction R.X. Rapport C.F.P., Bor- Brinkhuis, H., Visscher, H., 1994. New evidence for terrestrial deaux, ine¨dit. ecosystem collapse at the KT and Permian/Triassic. In: New Galeotti, S., Coccione, R., 2002. Changes in coiling direction Developments Regarding the KT Event and other Catastro- of Cibicidoides pseudoacutus across the Cretaceous-Tertiary phes in Earth History. LPI Contribution vol. 825, p. 17. transition of Tunisia: paleoecological and biostratigraphic Brinkhuis, H., Bujak, J.P., Smit, J., Versteegh, G.J.M., Vissch- implications. Palaeogeogr. Palaeoclimatol. Palaeoecol. er, H., 1998. Dino£agellate-based sea surface temperature Gaucher, G., 1981. Les Facteurs de la Pedogene©se. G. Lelotte, reconstruction across the Creaceous-Tertiary boundary. Pa- Dison, 730 pp. laeogeogr. Palaeoclimatol. Palaeoecol. 141, 67^83. Graham, D.W., Bender, M.L., Williams, D.F., Keigwin, L.D., Burollet, P.F., 1956. Contribution a l'Etude Stratigraphique de Jr., 1982. Strontium-calcium ratios in Cenozoic planktic foraminifera. Geochim. Cosmochim. Acta 46, 1281^1292. 31

Hallam, A., 1992. Phanerozoic Sea-Level Changes. Columbia taceous-Tertiary mass extinction in planktic foraminifera. University Press, New York, 266 pp. Palaeogeogr. Palaeoclimatol. Palaeoecol. Hansen, H.J., Toft, P., Mohabey, D.M., Sarkar, A., 1996. Klug, H.P., Alexander, L., 1974. X-ray Di¡raction Procedures Lameta age: Dating the main pulse of the Deccan traps for Polycrystalline and Amorphous Materials, 1st, 2nd edn. volcanism. Gondwana Geol. Mag. Nagpur India Spec. vol. Wiley, New York. 2, 365^374. Ku«bler, B., 1983. Dosage quantitatif des mine¨raux majeurs des Hansen, H.J., Toft, P., 2002. Magnetic susceptibility and Ni- roches se¨dimentaires par di¡raction X. Cahier de l'Institut rich spinels at the K-T boundary in El Kef, Elles and Ain de Ge¨ologie de Neuchaªtel, Se¨r. AX Nos. 1.1 and 1.2. Settara, Tunisia. Palaeogeogr. Palaeoclimatol. Palaeoecol. Ku«bler, B., 1987. Cristallinite de l'illite, methods normalisees Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of de preparations, methods normalisees de measures. Cahiers £uctuating sea levels since the Triassic. Science 235, 1156^ Institut of Geologie, Neuchatel, Ser. ADX 1, 13 pp. 1167. Li, L., Keller, G., 1998a. Maastrichtian climate, productivity Hess, J., Bender, M.l., Schilling, J.G., 1986. Evolution of the and faunal turnovers in planktic foraminifera in south At- ratio of strontium 87 to strontium 86 in sea water from lantic DSDP Site 525A and 21. Mar. Micropaleontol. 33, Cretaceous to present. Science 231, 979^984. 55^86. Karoui-Yakoub, N., Zaghbib-Turki, N., Keller, G., 2002. The Li, L., Keller, G., 1998b. Abrupt deep-sea warming at the end Cretaceous-Tertiary (K-T) mass extinction in planktic fora- of the Cretaceous. Geology 26, 995^998. minifera at Elles I and El Melah, Tunisia. Palaeogeogr. Pa- Li, L., Keller, G., 1998c. Diversi¢cation and extinction in laeoclimatol. Palaeoecol. Campanian-Maastrichtian planktic foraminifera of North- Kaiho, K. et al., 1999. Oceanic primary productivity and dis- western Tunisia. Eclogae Geol. Helv. 91, 75^102. solved oxygen levels at the Cretaceous-tertiary boundary: Li, L., Keller, G., Stinnesbeck, W., 1999. The late Campanian their decrease, subsequent warming and recovery. Paleocea- and Maastrichtian in northwestern Tunisia: palaeoenviron- nography 14, 511^524. mental inferences from lithology, macrofauna and benthic Keller, G., 1988a. Biotic turnover in benthic foraminifera foraminifera. Cretac. Res. 20, 231^252. across the Cretaceous/Tertiary boundary at El Kef, Tunisia. Li, L., Keller, G., Adatte, T., Stinnesbeck, W., 2000. Late Palaeogeogr. Palaeoclimatol. Palaeoecol. 66, 153^171. Cretaceous sea level changes in Tunisia: A multi-disciplinary Keller, G., 1988b. Extinction, survivorship and evolution of approach. Geol. Soc. Lond. 157, 447^458. planktic foraminifera across the Cretaceous/Tertiary bound- Lindinger, M., 1988. The Cretaceous/tertiary boundary at El ary at El Kef, Tunisia. Mar. Micropaleontol. 13, 239^263. Kef and Caravaca: Sedimentological, geochemical and clay Keller, G., 1993. The Cretaceous-Tertiary transition in the mineralogical aspects. PhD thesis, ETH, Zurich, 253 pp. Antarctic Ocean and its global implications. Mar. Micropa- Loutit, T.S., Hardenbohl, J., Vail, P.R., Baum, G., 1988. Con- leontol. 21, 1^45. densed sections: the key to age determination and correla- Keller, G., Benjamini, C., 1991. Paleoenvironment of the east- tion of continental margin sequences. In: Wilgus, C.K., Pos- ern Tehys in the early Paleocene. Palaios 6, 439^464. amentier, H., Ross, C.A., Kendall, C.G. (Eds.), Sea-Level Keller, G., Barrera, E., Schmitz, B., Mattson, E., 1993. Grad- Changes: An Integrated Approach. Soc. Econ. Paleontol. ual mass extinction, species survivorship and long-term en- Mineral. Spec. Publ. vol. 42, pp. 183^213. vironmental changes across the Cretaceous/Tertiary bound- Luciana, V., 2002. High-resolution planktonic foraminiferal ary in high latitudes. Geol. Soc. Am. Bull. 105, 979^997. analysis from the Cretaceous/Tertiary boundary at Ain Set- Keller, G., Li, L., MacLeod, N., 1995. The Cretaceous/Terti- tara (Tunisia): evidence of an extended mass extinction. Pa- ary boundary stratotype section at El Kef, Tunisia: How laeogeogr. Palaeoclimatol. Palaeoecol. catastrophic was the mass extinction? Palaeogeogr. Palae- MacLeod, N., Keller, G., 1991a. Hiatus distribution and mass oclimatol. Palaeoecol. 119, 221^254. extinction at the Cretaceous/Tertiary boundary. Geology 19, Keller, G., Li, L., in press. Evidence for progressive mass ex- 497^501. tinction and onset 500 k.y. before the the K-T boundary. MacLeod, N., Keller, G., 1991b. How complete are Creta- Dinofest 2000. ceous/Tertiary boundary sections? A chronostratigraphic es- Keller, G., Stinnesbeck, W., 1996. Sea-level changes, clastic timate based on graphic correlation. Geol. Soc. Am. Bull. deposits, and megatsunamis across the Cretaceous-Tertiary 103, 1439^1457. boundary. In: MacLeod, N., Keller, G. (Eds.), Cretaceous/ Martin, E.E., Macdougal, J.D., 1991. Seawater Sr isotopes at Tertiary Boundary Mass Extinction: Biotic and Environ- the Cretaceous-Tertiary boundary. Earth Planet. Sci. Lett. mental Changes. W.W. Norton, New York, pp. 415^450. 104, 166^180. Keller, G., Adatte, T., Stinnesbeck, W., Stuben, D., Kramar, Mc Arthur, J.M., 1994. Recent trends in strontium isotope U., Berner, Z., Li, L., Perch- Nielsen, K., 1998. The Creta- stratigraphy. Terra Nova 6, 331^358. ceous-Tertiary transition on the shallow Saharan platform McLean, D.M., 1985. Deccan Traps mantle degassing in the of southern Tunisa. Geobios 30, 951^975. terminal Cretaceous marine extinctions. Cretac. Res. 6, 235^ Keller, G., Adatte, T., Stinnesbeck, W., Luciani, V., Karoui, 239. N., Zaghbib-Turki, D., 2002. Paleobiogeography of the Cre- Meon, H., 1990. Palynologic studies of the Cretaceous/Tertiary boundary interval at El Kef outcrop, northwestern Tunisia: 32

Paleogeographic implication. Rev. Paleobot. Palynol. 65, Knight, J.D., 1984. Disruption of the terrestrial plant eco- 85^94. system at the Cretaceous-Tertiary boundary, Western Inte- Meon, H., Mannai, Armand, A., Desvignes, P., 2002. The K-T rior. Science 225, 1030^1032. boundary in El Kef section and some other sites (Tunisia): Saito, T., Yamanoi, T., Kaiho, K., 1986. End-Cretaceous dev- evolution of the continental vegetation. Palaeogeogr. Palae- astation of terrestrial £ora in the boreal Far east. Nature oclimatol. Palaeoecol. 323, 253^255. Monaco, A., Mear, Y., Murat, A., Fernandez, J.M., 1982. Sassi, M., 1974. La se¨dimentation phosphateë au Paleöce©ne Crite©res mineralogiques pour la reconnaissance des turbi- dans le sud et le centre ouest de la Tunisie. These Sci., dites ¢nes. C. R. Acad. Sci. Paris 295II, 43^46. Orsay, 292 pp. Moore, D., Reynolds, R., 1989. X-Ray-Di¡raction and the Schmitz, B., Keller, G., Stenvall, O., 1992. Stable isotope and Identi¢cation and Analysis of Clay-Minerals. Oxford Uni- foraminiferal changes across the Cretaceous/Tertiary bound- versity Press, Oxford, 332 pp. ary at Stevns Klint, Denmark: Arguments for long-term Orth, C.J., Gilmore, J.S., Knight, J.D., Pillmore, C.L., oceanic instability before and after bolide impact. Palaeo- Tschudt, R.H., Fasset, J.E., 1981. An iridium abundance geogr. Palaeoclimatol. Palaeoecol. 96, 233^260. anomaly at the palynological Cretaceous-Tertiary boundary Speijer, R.P., 1994. Extinction and recovery patterns in benthic in northern New Mexico. Science 214, 1314^1343. foraminiferal paleocommunities across the Cretaceous/Pale- Parsons, T.R., Takahashi, M., 1973. Biological Oceanographic ogene and the Paleogene/Eocene boundaries. Geol. Ultraiec- Processes. Pergamon Press, Oxford, 186 pp. tina vol. 124, 191 pp. Pospichal, J.J., 1994. Calcareous nannofossils at the K-T Stoll, H.M., Schrag, D.P., 1996. Evidence for glacial control of boundary; El Kef; no evidence for stepwise, gradual, or rapid sea level changes in the early Cretaceous. Science 272, sequential extinctions. Geology 22, 99^102. 1771^1774. Remane, J., Keller, G., Hardenbohl, J., Ben Haj Ali, M., 1999. Stinnesbeck, W., Keller, G., Adatte, T., Lopez-Oliva, J.G., International workshop on Cretaceous-Paleogene transitions MacLeod, N., 1996. Cretaceous-Tertiary boundary clastic in Tunisia: The El Kef stratotype for the Cretaceous-Pale- deposits in Northeastern Mexico: Impact Tsunami or sea- ogene boundary recon¢rmed. Episodes 22, 47^48. level lowstand. In: MacLeod, N., Keller, G. (Eds.), Creta- Renard, M., 1986. Pelagic carbonate chemostratigraphy (Sr, ceous/Tertiary Boundary Mass Extinction: Biotic and Envi- Mg, N18ON13C). Mar. Micropaleontol. 10, 117^164. ronmental Changes. W.W. Norton, New York, pp. 471^517. Robaszynski, F., Gale, A., Juignet, P., Amedro, F., Harden- Urrutia-Fucugauchi, J., Ortega-Ramirez, J., Cruz Gatica, R., bohl, J., 1998. Sequence stratigraphy in the upper Creta- 1997. Rock-magnetic study of late Pleistocene-Holocene ceous of the Anglo-Paris Basin: Exempli¢ed by the Ceno- sediments from the Bab|¨cora lacustrine basin, Chihuahua, manian stage. In: Mesozoic and Cenozoic Sequence northern Mexico. Geo¢s. Int. 36, 23^47. Stratigraphy of European Basins, SEPM Spec. Publ. vol. Vincent, S.J., Macdonald, D.I.M., Gutterridge, P., 1998. Se- 60, pp. 364^386. quence stratigraphy. In: Doyle, P., Bennet, M.R. (Eds.), Robert, C., Chamley, H., 1991. Development of early Eocene Unlocking the Stratigraphical Record: Advances in Modern warm climates as inferred from clay mineral variations in Stratigraphy. oceanic sediments. Glob. Planet. Change 89, 315^332. Vonhof, H.B., Smit, J., 1997. High resolution late Maastrich- Robert, C., Kennett, J.P., 1992. Paleocene and Eocene kaolin- tian-early danian oceanic 87Sr/86Sr record: implications for ite distribution in the South Atlantic and Southern Ocean Cretaceous-tertiary boundary events. Geology 25, 347^350. Antarctic climate and paleoceanographic implications. Mar. Weaver, C.E., 1989. Clays, muds and shales. In: Development Geol. 103, 99^110. in Sedimentology vol. 44. Elsevier, Amsterdam, 819 pp. Said, R., 1978. Etude stratigraphique et micropaleöntologique Wolbach, W.S., Lewis, R.S., Anders, E., 1985. Cretaceous ex- du passage Cre¨tace¨-Tertiaire du synclinal d'Elles (Re¨gion tinctions: evidence for wild¢res and search for meteoritic Siliana-Sers) Tunisie Centrale. The©se 3ie©me Cycle, Universite¨ material. Science 230, 167^170. Pierre et Marie Curie, Paris IV, 275 pp. Tschudy, R.H., Pillmore, C.L., Orth, C.J., Gilmore, J.S.,